Seasonal hotspots of leatherback turtle bycatch in the Atlantic Ocean: Insights from longline observer data

Bycatch from pelagic longline fisheries poses a serious threat to the endangered leatherback turtle in the Atlantic Ocean. However, the spatiotemporal distribution of bycatch risk remains poorly understood, largely due to data limitations such as sparse observer coverage, zero-inflation, and inconsistent temporal sampling. Here, we analyze 18 years (2002–2019) of Japanese longline observer data to identify seasonal high-risk areas for turtle bycatch, using a zero-inflated binomial model based on stochastic partial differential equations (SPDE) combined with hotspot analysis. This framework allows us to extract meaningful spatial patterns from data-poor situations and generate spatially explicit estimates of relative leatherback density and bycatch risk. Our results reveal that bycatch hotspots occur predominantly near the African coast in the first quarter and expand across both the African coast and the broader North Atlantic in the fourth quarter. Seasonal differences in risk were more pronounced than interannual fluctuations, aligning with known migratory behaviors of leatherbacks. These findings underscore the importance of season-specific conservation strategies such as time-area closures or dynamic bycatch avoidance measures, providing actionable spatial and seasonal risk maps that could inform the design and timing of mitigation measures. More broadly, our approach offers a practical solution for assessing risk in other threatened marine taxa under data-limited conditions and enhances evidence-based conservation planning in marine ecosystems.